Glass
Transition region
Traditional presentation of relaxation map in
polymers includes 4 regions:
Log G(t)
Segmental
Relaxation -Glass-like behavior;;
-Transition region;
Rubber -Rubbery plateau;
Rouse
Modes Flow -Terminal relaxation (flow).
Reptation
0 1 2 3 4 5 6 7 8 9
10 10 10 10 10 10 10 10 10 10
Log t
Secondary relaxations (
)
Segmental relaxation
(-relaxation) However, there are a few relaxation processes
Fast
even in a glassy state (or at time scales shorter
G(t)
picosecond
Log G
relaxation th segmental
than t l relaxation).
l ti )
In particular, secondary relaxations influence
mechanical properties of polymers.
0 1 2 3 4 5 6 7 8 9 10 11 12 13
10 10 10 10 10 10 10 10 10 10 10 10 10 10
Log t
Frequency map of polymer dynamics
Chain dynamics:
y Rouse and terminal relaxation
Structural relaxation, -process
Vibrations
Secondary relaxations Fast relaxation FREQUENCY,
Spin-Echo
Neutron Scattering, S(Q,) Back-sc. Time-of-Flight
Secondary relaxations ()
Segmental relaxation There are many relaxation processes in polymeric
(-relaxation) systems. They appear with different characteristic
Fast times ,
relaxation times, or frequencies.
frequencies The names of
Log G(t)
picosecond
the processes (-, -, -) reflects the order in
relaxation
which the processes appear (excluding chain
relaxation). There is no physical meaning behind
these names. However the -process
names However, process is usually
assigned to the segmental relaxation.
0 1 2 3 4 5 6 7 8 9 10 11 12 13
10 10 10 10 10 10 10 10 10 10 10 10 10 10
Log t
Different experimental techniques should be used
S e c o n d a r y r e la x a t io n s ( ) for measurements of the relaxation spectrum in so
S e g m e n t a l r e la x a t io n
( - r e la x a t io n ) broad time or frequency range. Another way can
g G ()
Secondary
S d relaxations
l i usually
ll show
h A h i temperature dependence
Arrhenius d d with
i h an activation
i i energy
V~20 kJ/mol. Segmental relaxation has strongly non-Arrhenius temperature dependence, i.e. its
apparent activation energy is a function of temperature.
In reality there are many corrections to so simple picture. It was found that the shift factor for various
polymeric systems is usually well described using Williams-Landel-Ferry (WLF) equation:
C1 (T TS )
logg a T
C 2 (T TS )
An example of storage compliance
data for poly-n-octyl methacrylate.
Composite curve obtained by
plotting the data with suitable shift
factors Original data were extended
factors.
over 2 decades only. The combined
curve covers 11 orders in frequency.
The time-temperature equivalence (or superposition) is used by many researchers for analysis of
viscoelastic properties of polymers. You find it in all textbooks. It explicitly assumes that all
viscoelastic processes have the same temperature variations. One should keep in mind, however, that in
most cases the time-temperature superposition breaks down. It will be discussed later.
Segmental Relaxation -process
The main
Th i structural
t t l relaxation
l ti ini polymers
l i the
is th segmental, ll d -relaxation.
t l so-called l ti It controls
t l diffusion
diff i
and viscosity, rotation of monomers. The -process is the relaxation on time scale shorter than the
rubbery plateau and the Rouse modes. It is responsible for the transition region and is associated with
the glass transition.
It is usually ascribed to micro-Brownian motion of chain segments. Most authors agree that the -
process is related to conformational changes (like gauche trans transition). Because the glass transition
is directly related to the -relaxation, both, Tg and segmental relaxation (the -process), show the same
dependence on Mw and crosslinking.
crosslinking
0.8
og G"()
G(t)
H()
0.6
-1
~
appears as a Lorentzian in the
~
0.4 frequency
q y domain and
Lo
1/
1/e
0.2
corresponds to a single
0.0
Time Log Log
0.8
-b
Log G"()
G(t)~exp(-(t/) )
~ H() is a distribution
G(t)
H()
~
0.6
0.4
1/e
of , i.e. it
characterizes how
0.2
00
0.0
many relaxators have
Log
Time Log relaxation time .
The same temperature dependence in non-polymeric systems is usually described by the Vogel-
Fulcher-Tammann (VFT) equation: B The equations are equivalent: C2=T*-T0;
0 exp
T T0 2.3C1C2=B, C1(T*)=log[(T*)/0].
It seems that different processes have similar temperature dependence at high temperatures only
[Roland, et al. Macromolecules 34, 6159 (2001)].
Breakdown of time-temperature equivalence
appears also in the temperature dependence of the
rubbery plateau in entangled polymers.
The reason for the breakdown of the time-temperature equivalence remains the subject of discussions.
First of all, segmental relaxation is driven by energetic forces, while Rouse modes are entropic and
reptation is controlled by a disenteglement time.
time Temperature dependence in the Rouse and reptation
models is introduced through monomeric friction coefficient. Temperature variations of the friction
coefficient may be different from temperature dependence of segmental relaxation.
Reptation model assumes that the same friction coefficient is involved in the Rouse modes at short
times (inside the tube) and in the reptation modes at long times. The data on PVAc show that the
friction coefficient might be different. Ferry and co-workers proposed [J.Colloid Sci. 14, 135(1959) and 17, 10
(1962)] that a number of entanglements can be a function of temperature. It is not clear whether this is
really the case.
Thus, one should be aware that time-temperature equivalence (superposition), although very often used
and presented in all textbooks, is oversimplification that fails for many polymers.
Concluding Remarks:
1. Time-temperature superposition assumes that all relaxation processes have similar temperature
dependence. It seems to be correct at high T, but it fails when temperature approaches Tg.
2. Spectral
p shape
p of segmental
g relaxation is always
y stretched ((non-exponential).
p ) Stretchingg of the
relaxation spectrum is a characteristic feature for relaxations in complex systems. The
mechanism of the stretching (dynamic heterogeneity or intrinsically non-exponential process)
remains unclear, although there are many indications of heterogeneous dynamics at time scale
smaller than segmental
g relaxation time.
Temperature variations of volume. Tg depends on Change in specific heat at temperatures around Tg.
cooling rate.
However, Tg is an ill-defined quantity. It depends on a cooling or heating rate. Due to that reason also
another definition of Tg is accepted: Tg is a temperature where segmental relaxation time ~100 sec.
There is no phase transition of any kind at Tg.
The nature of the glass transition phenomenon remains a subject of discussions. However, the basic
event of the observed transition from a liquid to a solid is a kinetic phenomenon. Glass transition is a
freezing of segmental relaxation.
Glass transition depends on pressure: increase in
P leads to increase in Tg of polymers. dTg/dP is
diff
different
t for
f different
diff t polymers.
l
Thus we have an equation similar to WLF equation, with T0=Tg, Tg, C1=b/2.3f
b/2.3fg; C2=ffg/af. In that case one
can relate the model parameters and WLF parameters: fg=b/2.3C1 and af=b/2.3C1C2. However, because
the constant b is an arbitrary parameter, no direct estimates of the free volume fraction can be obtained.
f A simple picture behind the free volume approach is based on the
assumption that an empty (free) volume is needed for molecular
V motion.
ti Thi free
This f volume
l d
decreases with
ith decrease
d i T.
in T That
Th t leads
l d
Vg fg Vo to slow down of the motion. Relaxation time at Tg crosses the
Vg experimental time scale and freezing of the structure (including
free volume) occurs. It gives frozen free volume fg. If one would
T Tg T cooll the
th sample l down
d with
ith infinitely
i fi it l slow
l rate,
t fg=0
f 0 willill be
b
reached at T that would be equivalent to T0 of VFT equation.
When one considers b~1, the value for the free-volume fraction at Tg for many polymers falls in the
range
g fg~0.013-0.034. WLF pproposed
p a universal value fg~0.025. Later,, another relation was also
suggested: fg=10-4*Tg+0.07 [Boyer, Simha, J.Polym.Sci.Polym.Lett. 11, 33 (1973)]. It was further modified: fg is not
frozen at Tg; one should distinguish fractional empty free volume or dynamic free volume from total
free volume.
Free volume approach has been extended to include pressure effects:
b T T0 ( P )
log aT (4)
2.3 f 0 f 0 / f T Tg ( P )
Here f0 is f at T=T0. (P) is a function that depends on the pressure-dependent coefficient of thermal
expansion of the free volume, f(P). It has been shown that the equation 4 describes reasonably well T-
and P- dependencies for some polymers.
Nevertheless, the free-volume approach has been criticized for many problems. In some cases
unreasonable parameters of the free volume should be assumed in order to describe data for some
materials. It has been also demonstrated that holding free volume constant (by varying simultaneously P
and T) leads to different viscosity, suggesting that not only density, but also temperature play role in the
glass transition.
The influence of molecular weight on Tg, chain-end free volume approach
Tg for many polymers depends on molecular weight Mn. At not very low Mn, Fox-Flory empirical
equation describes reasonably well the molecular weight dependence of Tg:
K
Tg ( M n ) Tg () (5)
Mn
pp
Free volume approach explains
p the Fox-Floryy equation,
q assumingg that chain ends contribute an excess
free volume. In that case, decrease in Mn leads to increase of chain ends concentration and increase of
free volume. Increase in free volume leads to decrease in Tg.
Lets assume that chain end has a free volume . Then the free volume per unit volume is 2NA/Mn,
where is density,
y NA is Avogadros
g constant. Assumingg that fgg is independent
p of Mn, the excess free
volume introduced by chain ends should be compensated by the thermal contraction:
2N A
2N A / M n f [Tg () Tg ( M n )] Then Tg ( M n ) Tg () (6)
f Mn
The Fox-Floryy equation
q A/fMn.
is obtained with the constant K=2N
380
340
220
3 4 5 6
10 10 10 10
Mn
Thermodynamic Approach
Any conformational
A f ti l changes
h require
i some cooperative
ti motion
ti off
a few molecules (cooperative domain). Domain consist of z
conformers, each has c1 number of states. The conformational
entropy for 1 mole of conformers in which there are Nz domains
consisting of z conformers: S c N z k ln c1
N A N A k ln c1 s *
z
Nz Sc Sc
here ss* is the conformational entropy of one mole of conformers
where each conformer relaxes independently. At high T>T* when
there is no cooperativity s*=Sc. At T<T* Sc drops faster than s*,
assumption
T * T T0 C
Sc s*; 0 exp
T * T0 T c
TS
An equation
q is the energy
for relaxation time (( gy barrier for one conformer to relax):
)
z 1 s * 1 T T *
ln / *
k T T * k TS c T *
(7)
kT * T T0
Thee lattice
a ce modelode ppredicts
ed c s thee eexistence
s e ce oof a true
ue seco
second-order
d o de
transition at a temperature T2. The number of allowed arrangements of
the molecules decreases with decreasing T because: (i) the number of
holes decreases; (ii) the configurational entropy of the molecules
deceases because the chains favor low-energy gy states at lower T.The
T(P) transition line defines the point T2(P) where the total
configurational entropy first becomes 0. In that respect, the T(P) line
represents the thermodynamic glass transition in experiments of long
time-scale.
Gibbs DiMarzio theory gives rather complicated prediction for the molecular weight dependence of
Tg
g [McKenna, Compreh.Polym.Sci. 2,311 (1989)]]:
2 exp
x ln v0 1 v0 ( x 1)(1 v0 ) ln 3( x 1) kTg kT
ln 1 g
ln 1 2 exp (8)
x 3 1 v0 1 v 0
1 2 exp 2kTg
2 xv0 x
kT
g
Here x is twice the degree of polymerization and v0 is the volume fraction of holes, Tg is the glass
transition temperature T2. The eq.8 describes well Tg(Mn). It even describes the deviation from the Fox-
Flory equation at smaller Mn.
The theory also makes an interesting prediction for ring polymers. It has been observed that Tg increases
with decrease in Mn in ring PDMS. This behavior has been described qualitatively using the eq.8.
PVC PDMS
rings
linear
Tg vs 1/M in PVC. Solid line shows Fox-Flory relationship. Tg in ring and linear PDMS compared to predictions
The dashed lines shows Gibbs DiMarzio model predictions of Gibbs-DiMarzio theory [from Guttman, DiMarzio,
(eq.8) [from Pezzin, et al. Eur.Polym.J. 6, 1053(1970)]. Macromolecules, 1988]
Influence of Molecular Structure and Architecture on Tg
Flexibility of the main-chain. Flexible group (for example, ether link) will enhance main-chain
flexibility and reduce Tg. Inflexible group (for example, terephthalate) will increase Tg.
Si-O-Si is a very flexible link. As a result, PDMS has one of the lowest Tg known for polymers.
Increasing the length of side groups reduces Tg, mostly due to increase in free volume.
Role of tacticity and microstructure. Tacticity of polymer chain can affect flexibility of the chain,
packing of chains (free volume) and in this way will influence glass transition temperature. Tacticity
strongly
l affects
ff crystallization
lli i off polymers.
l
C C
C C 1,2-PB, Tg~270K
-C C-
Tg in polybutadienes is a smooth function of vinyl content [Hofmann,
H f
et al. Macromolecules 29, 129 (1996)]. This is an example of Tg in a
random co-polymer.
150
350 It is important
p to emphasize
p
K]
Tg [K
140
Me that the molecular weight
Tg [K]
180 180
174 174
Linear Thus influence of molecular architecture on Tg and
Star segmental dynamics remains unclear and
172 172
0 1 2 3 4 5 0 2 4 6 8 10 experimental data are controversial.
4 4
10 /Mn,tot 10 f/Mn,tot
320
Validity of the chain-end free volume
approach
pp even for dendritic ppolymers
y was
proposed in [Wooley, et al. Macromolecules 26, 1514 300
(1993)]. Tg in dendritic poly(benzyl ethers)
Tg [K]
with various microstructures scales with 1/M.
However, it appears
pp that it scales better with 280
Mtot than with f/Mtot.
[G]-OH
260
[G]2-[C]
[G]3-[C]
0 5 10 15 20 25 30 50 60 70 80 90
4 4
10 /Mtot 10 f/Mtot
-naphthyl salicylate
phenyl salicialte
tricresyl
phosphate Tg varies non-linear with concentration of solvents. One of the
methyl explanation is based on the free-volume approach. Fractional free
acetate volume of the solvent fs is higher than that of the polymer fp (solvent
benzene has lower Tg). Assuming that the fractional free volumes are not
nitrobenzene additive, the total fractional free volume:
f vs f s v p f p k v vs v p (11)
Here vs and vp are volume fractions, k is a negative constant ~10-2.
Eq.11 assumes that occupied volumes are additive and that agrees
with measurements of density. Assuming iso-free-volume state at Tg
the following dependence is predicted [McKenna]:
vs sTgs v p pTgp k v vs v p (12)
Tg ( vs )
vs s v p p
Here is the thermal expansion coefficient of the fractional free volume. The eq.12 describes the data
well with reasonable parameters.
Free volume ideas were also used for description of Tg in co-
polymers. Additivity of volumes of the two components is assumed Poly(butadiene-co-styrene)
] T Tg vR w1vR1 w2 vR 2
[Gordon, Taylor, J.Appl.Chem. 2, 493 (1952)]:
T Tg vG w1vG1 w2 vG 2
Here vR and vG are the specific volumes of the copolymers in the rubbery
and glassy states,
states w is the weight fraction.
fraction The authors predict for Tg:
kw T w2Tg 2
Tg ( w2 ) 1 g 1 (13)
w1 kw2
Here k=2/1, is a difference of the thermal expansion coefficient
in the rubbery and in the glassy state. The eq. 13 describes the data for
poly(butadiene-co-styrene) reasonably well. Note that the eq.13 is
identical to the eq.(12) with kv=0.
There are a few models proposed for the description of Tg in miscible polymer blends. One of the recent
ideas proposed by Lodge and McLeish [Macromolecules 33, 5278 (2000)] takes into account effective
concentration of monomers A and B around a typical monomer A: eff S (1 S )
Where is the volume fraction of A component and S is the self-
C m
concentration term: S 0 (14)
n0 N AV
Here m0 is mass of a monomer, n0 is the number of backbone bonds per
monomer, is the density, V is a volume that influences the relaxation of
a monomer. It is assumed to be V~lK3 (lK is the Kuhn length).The main
idea is that the monomer A experience higher concentration of A because
of its connectivity. As a result, you have a distribution of eff. The model
explains many characteristic properties of relaxation and Tg in miscible
blends, even gives good quantitative predictions for some systems.
However, it fails for some other systems.
Glass transition in thin polymer films: Developments in nano science and nano technology
leads to decrease of characteristic size of elements. Qualitative difference appears when we
approach length scales ~5-50 nm.
One example of that is variation of Tg in thin polymer films [review by Forrest, Dalnoki-Veress, Adv.Coll.Interf.Sci.
94, 167 (2001)]. It has been observed that Tg of polymer films drops with decrease in film thickness h. The
best ppolymer
y analyzed
y is PS. Various techniques
q show dropp of Tgg ~30-50K when h ~10 nm. Dropp of Tgg
depends on substrate, suggesting importance of a polymer-substrate interactions. In particular, it has been
shown that for polymer physically grafted to substrate Tg can increase with decrease in h.
Another example: PMMA on Au -> Tg decreases with decrease in h, while PMMA on SiO2 -> Tg
increases with decrease in h.
No significant dependence of the effect on Mw has been observed in thin supported films.
films Tg(h) is
usually described by an equation: a
Tg ( h ) Tg 1
bulk
h
Even stronger effect has been reported for thin free
standing films (no substrate). In that case, strong
d
dependence
d off the
th effect
ff t on molecular
l l weight
i ht has
h
been observed. There are two regimes: lower Mw
(for PS~120,000 370,000), and higher Mw
(>370,000).
Eff t att lower
Effect l Mw is i similar
i il tot the
th effect
ff t in i
supported films but is ~2 time stronger.
Much stronger dependence appears at higher Mw, it
starts to deviate from the bulk behavior at h~REE
( dt
(end-to-endd distance).
di t )
Traditional picture for these variations is related to a surface layer that is assumed to have higher
mobility than the bulk polymer.
polymer Of course,
course if it interacts strongly with a substrate,
substrate it might have lower
mobility. The thickness of the layer might be different for different polymers and for PS is estimated to
be ~3.5-4 nm. Decreasing the thickness of the film enhances the influence of the layer on the film
properties.
Surface layer
y
Bulk properties Film
Surface layer
2. There are two main approaches for description of the glass transition: free volume and
thermodynamic (Gibbs-DiMarzio). Both describe the phenomena qualitatively and both have
various problems.
problems
3. Chemical structure, chain length and architecture of macromolecules affect their glass
transition temperature.